Quote:
Originally Posted by CherryCreek
Here's an interesting data point from CREJ that I find somewhat shocking, and contrary to many of the broad discussion points on this board for years:
While more residents began renting, their share decreased. While the number of Denver renters went up by 17% in the past decade, the number of homeowners increased faster, by 21% since 2010. The share of Denverites who rent shrunk by 2% since 2010, down to 45% at the end of the decade.
The figures come in contrast with growth trends on a country level – the renter population grew two times faster than the owner population in the past decade, and the tenant share expanded by 3%, up to 34% in 2018.
https://crej.com/news/accelerated-gr...s-past-decade/
Can this be right ? Despite all the apartment construction, the population boom, and the general sense that both SFH and condo construction has not kept up with Denver's rapid growth, the percentage of persons who own vs. rent has gone UP in Denver in the last decade???
I'm really surprised. (though interestingly, the share of renters (45%) in Denver is much higher than nationally (34%), but per CREJ, it was that way here BEFORE the boom).
|
First, I would encourage everyone to always view these types of articles with a critical eye. The article appeared in CREJ but the author is from RENTCafe, a service by Yardi, a real estate services firm, that produces click-bait-y articles. The author of the article currently lives in Romania, according to her LinkedIn profile, and her job position at Yardi is listed as "Content Marketing Specialist." Right there, this should give you pause.
Next, the article contains not a single worthy citation. Embedded within the graphics are citations of "Source: U.S. Census" but that is not sufficient. The amount of data available through the Census Bureau is staggering, and a proper citation is one that should allow the reader to go directly to the source of the data. The dozens of numbers and percentages that are tossed around in the article text have no source references provided at all, except for the paragraph about rent increases, and that citation references one of RENTCafe's previous articles. The author also uses sloppy and imprecise language throughout, a definite warning sign when reading something that throws a bunch of numbers and statistics at you.
The section of the article that seems to have piqued everyone's interest here is this one:
"While more residents began renting, their share decreased. While the number of Denver renters went up by 17% in the past decade, the number of homeowners increased faster, by 21% since 2010. The share of Denverites who rent shrunk by 2% since 2010, down to 45% at the end of the decade."
Because there were no source references other than the generic "U.S.Census Bureau," I did some digging around and found the exact data tables the author used, so let's explore this a bit more.
(Before I begin, note the phrase "the number of homeowners increased faster..." This is wrong. The number didn't increase
faster, the number
increased by a greater percentage. This is an example of the author's sloppy use of data-related language.)
For this section of the article, the author used for the 2010 data "Table ID B25008 - Total Population in Occupied Housing Units" from the 2010 ACS 1-Year Estimates data set. For the 2018 data, the author used "Table ID B25008 - Total Population in Occupied Housing Units" from the 2018 ACS 1-Year Estimates data set. The ACS (American Community Survey) data are estimates, based on a sample the Census Bureau takes over the course of the year. As estimates, they include a Margin of Error.
Right off the bat, we have a problem: 1.) The data used are estimates with margins of error, not actual counts. The author never mentioned that in the article. 2.) The data samples were collected over the course of 2018, so the author's frequent use of "this decade" and "during the decade" is not quite correct. This is a comparison of 2010 vs. 2018, not a comparison of the entire decade through the end of 2019.
Here are the 2010 and 2018 data the author used. I obtained this from the above-referenced data sets on the Census Bureau's American Community Survey
website and formatted them into these exhibits:
Please note the huge Margins of Error. This is what you get when you use ACS 1-Year data. There are also 5-Year data sets available that have lower margins of error, but the data isn't as current since it is a blend of samples taken over a five-year period.
The other thing I'll note before going on is that we have sample data with margins of error at both ends (2010 and 2018), giving us "double margins of error" comparisons. What we do have for the 2010 data is the actual Census counts for these populations from the 2010 Decennial Census. Using that for the 2010 data might have been better since it would have eliminated the margin of error at one end. I looked up this data from the 2010 Census (Table ID: H11) and found the actual "Owner Occupied" count to be 309,975 and the actual "Renter Occupied" count to be 274,202. So right here we can see that if the author had used the 2010 Census counts instead of the 2010 ACS 1-Year estimate (with its high MoE), the percentages would have been 53% Owner Occupied / 47% Renter Occupied instead of the 54% / 46% split using the ACS data.
But let's stick with the 2010 ACS estimate data since that's what the author used, and let's assume the 2010 ACS estimates were true counts, with no margin of error. This locks in the 2010 end and leaves us with only a margin of error on the 2018 side.
As you can see from the table above, the MoE for the total population in occupied housing units is relatively low, only +/- 2,294, since estimating the city's housed population is a fairly established methodology. However, the
distribution of that population between owner-occupied and renter-occupied units is another thing, with huge margins of error in the +/- 11,000 range.
The Census Bureau says "the value shown here is the 90 percent margin of error. The margin of error can be interpreted roughly as providing a 90 percent probability that the interval defined by the estimate minus the margin of error and the estimate plus the margin of error (the lower and upper confidence bounds) contains the true value."
So what we could do is add and subtract 11,222 to/from the Owner Occupied number of 382,650 to explore the extremes for this number on the high and low end (while still maintaining 90% confidence), and we could add and subtract 11,175 to/from the Renter Occupied number of 318,859 to do the same. I did that and then calculated the % Total for each as well as the Difference from the 2010 numbers and the % Change:
The area in yellow is the same as the 2018 section in the first table. These are the 2018 estimates. The area in orange is where I subtracted the MoE from the Owner Occupied number and added the MoE to the Renter Occupied number to give us the extreme "Owner Low / Renter High" scenario. The area in blue is where I subtracted the MoE from the Renter Occupied number and added the MoE to the Owner Occupied number to give us the extreme "Owner High / Renter Low" scenario.
What this show us is that, using the same data that the author used and by understanding we are dealing with margins of error, we could craft three completely different narratives. The first one is the narrative the author used in the article (after cleaning up her sloppy language and fixing her subtraction error between the 2010 and 2018 percentages for Renter Occupied). The remaining two are the narratives we would have under the two "margin of error" scenarios:
Estimates Narrative: "While the number of people living in Renter Occupied units in Denver went up by 17% from 2010 to 2018, the number of people living in Owner Occupied units went up by 21%. The share of Denverites who rent decreased by 1 percentage point, from 46% in 2010 to 45% in 2018."
Owner Low/Renter High Narrative: "While the number of people living in Owner Occupied units in Denver went up by 18% from 2010 to 2018, the number of people living in Renter Occupied units went up by 21%. The share of Denverites who rent increased by 1 percentage point, from 46% in 2010 to 47% in 2018."
Owner High/Renter Low Narrative: "While the number of people living in Renter Occupied units in Denver went up by 13% from 2010 to 2018, the number of people living in Owner Occupied units went up by 25%. The share of Denverites who rent decreased by 2 percentage points, from 46% in 2010 to 44% in 2018."
Imagine the headlines we could spin depending on which of the three scenarios we chose to go with! The fact that we have such huge margins of error that can completely change the narrative, and that this was ignored entirely by the author, is unforgivable.
Also note that the author's use of "number of renters" and "number of homeowners" isn't really correct. It is "people living in renter-occupied units" and "people living in owner-occupied units" which includes children and everyone else---not just the persons on the lease or on the property deed. The author also confused percent with percentage points, a serious issue. The "share of Denverites who rent shrunk by 2% since 2010" is wrong. What the author meant was "the share of Denverites who rent shrunk by 2 percentage points since 2010" (and, as noted, the figure itself is incorrect--it is 1 percentage point, not 2).
All of this from just one section of the article. Makes you question everything else in the article, right?